Revision: V2 26.10.2001
This consisted of a series of presentations showing status of recent sites work and plans for the future, concentrating on the four proposed large telescope projects (CELT, Cornell-Texas, GSMT, OWL).
The presentations were as follows:
We moved to the main topic for the workshop - characterization of sites in terms of the atmospheric parameters. The list below is in rough chronological order, but I have in various places included discussions and comments made at other times when it seemed sensible to do so. Various editorial comments are usually but not always in brackets...
The other site of immediate interest in Mauna Kea. This is gridded at 1-m resolution. The ELT site is identified on the MK Master Plan but it may not be as good as the sites where present telescopes are placed. But in principal these and other sites could be looked at. An issue is that Dave De Y.is very busy - He is President of the Aspen Center for Physics and Project Scientist for the National Virtual Observatory project. Funding a post-doc skilled in computational fluid dynamics to work on these projects under Dave's supervision would seem a priority if this work is to be advanced rapidly.
CELT have a fast-track schedule, albeit relying on the arrival of funding for CoD studies and then for the facility. There is pressure therefore to characterize a prime site, nominally MK, and also one or more backups, likely to be in the Western Hemisphere. The scientific and technical drivers on site selection will become more clear with publication of the "Green Book" -- the (white) pages for the "sites" chapter were distributed late in the meeting by Matthias Schoeck; Jerry N. and Mike Bolte told us that the mid-IR is not of much interest, the near-IR is, and the optical also is important. At first order, this would rule out ultra-high altitude sites selected for lower PWV as opposed to low cloud (the trade in Chile), and would also indicate that low light pollution is important (thus Tolar, near to Tocapillo is probably ruled out despite being spectacularly clear).
Matthias S. and Jerry N .espoused the use of Acoustic Sounders (SODAR) for determining the extent of the ground layer, despite some dismissive remarks that they were uncalibratable, worked poorly in dry conditions typical of astronomical sites, and S/N was degraded by wind-induced noise. Andrei T. mentioned the positive experience of Walter & Gur'yanov, and although quantitative calibration is difficult, the proposed use on MK is differential so it may be a good tool. Alistair Walker gave Matthias S. some docs from Atmospheric Research (ARPL) in Australia who appear to be (the) experts in the manufacture of SODARs. Note their WWW pages show SODARS sited in front of New Zealand rain forest, and people in full wet-weather gear... Andre E. then mentioned his measurements several years ago using 30-m towers with microthermal sensors, which included positions on the summit ridge and the plateau. Although the structure of a 30-m telescope is huge and the mirror may nominally be several 10's of meters off the ground, it's not always pointing at the zenith, and might easily dip into the ground layer depending on its thickness. Andre E. said that on the plateau at 30-m you were still in the ground layer.
Marc S. yesterday showed a table of max. horizontal acceleration statistically expected in 50-years from seismic activity, for a number of sites. There was some doubt whether the very low figure for MK was really correct, and Marc S. will investigate. And do the figures take into the account the damping (or amplification...) that may be a function of the type of rock that the telescope structure is bolted to at the specific site?
There was some argument about the importance of doing this analysis for our potential sites, since given the low vertical resolution one was not going to be able to calculate CN**2 profiles with much resolution, in particular resolution near the ground would be very poor. Andrei Tokovinin thought that since the value was in what it told you about turbulence in the free atmosphere, and it will be some years before we have (profilometer) measurements to compare with, what was the use?
Elena M. also had doubts on the use of the analysis. It is known that there exists a "qualitative"correlation between the meteorological parameters and the CN2 but it is not possible to establish a 1-1 correlation. If this would be possible, there would be no need to run meso-scale models. The CN2 depends not only on the temperature and the wind intensity but also on other parameters (like the mixing length) that are not meterological parameters and cannot be retrived it by the re-analysis. The mixing length represents the way for us to describe in a macroscopic way the microscopic spatio-temporal fluctuations. She went on to say she thought that the solution of the NCEP re-analysis is not physically correct; the problem is not that we have a CN2 with a low resolution, the problem is that we can not correctly reconstruct the CN2. It is possible to estimate the shear of the wind and the gradient of the temperature at the altitude of the jet stream but it is not possible to find a precise correlation with CN2.
Andre E. countered that the main value would be to follow climatology changes over a long time base, and not to use it for a CN**2 profile comparison. Weather patterns that cause increased turbulence (such as the southward movement of the Bolivian circulation) could be characterized. All good sites are in the desert-type climate which is intrinsically unstable, highly influenced by global change (less true for MK though). He also pointed out that the exploitation of this dataset needed a pilot study, and suggested testing over the 14 years for which there was Paranal seeing data. That then provoked a discussion of whether there would be a correlation, given the lack of information at low altitude. AE countered that you should predict the high altitude (jet stream boundaries) turbulence always, and if the ground wind tracked the free atmosphere wind then the turbulence layer at 1-3 km should be missing too, and that should be detectable from the NCEP data and verifiable by ground measurements. So, as Marc S. pointed out, comparison with ground weather stations would be important, since comparing local (T,V) with global (NCEP) gives a diagnostic of the presence or absence of local effects. And clearly the long-term information from an NCEP re-analysis would be very valuable. He further noted that all good sites have flat terrain upwind (terrain can also = ocean).
Alistair W suggested to Andre E that he expand his few- sentence justification of the NCEP re-analysis (as part of a proposal to CELT & NIO) to 2-3 pages and it would be circulated to the group for comment and further discussion.
Andre E also mentioned that a follow-up project would be to use the NCEP re-analysis output as the input to a model such as MM5 for a detailed analysis of a few sites.
UH provides http://hokukea.soest.hawaii.edu/forecast/models/mm5 CN**2 profiles using the mesoscale model MM5 as part of their weather predictions. Although there is qualitative agreement between the integrated profiles and the telescope seeing, there has been little (no?) serious attempt to compare the results with the SCIDAR runs for instance. Andre E commented that it is very important to do these comparisons since past such comparisons elsewhere had not shown much success. Unfortunately the person responsible for the code has left UH, Richard Wainscoat reported that there is a new post-doc taking over, she starts in November. The code and methodology should be recovered as a high priority initial activity.
Elena M. is using a different mesoscale atmospheric model (Meso-NH), and has applied it at several Observatory sites (see http://www.astroscu.unam.mx/~elena/. She countered the criticism above by saying that during the past four years there has been much progress, as documented in the literature (see her WWW-site), and in particular the recent paper Masciadri & Jabouille (2001) . She also could do with post-doc level help, and a specific near-term activity would be to compare her predictions with measurements, ie to calibrate the models, probably at Paranal and MK. There were some discussions off-line with Andrei T, Marc S and Matthias S to lay out a plan.
SCIDAR runs are expensive and require what used to be called a medium-aperture telescope. Andrei T. reported that two Profilometers are being built, for CTIO and ESO. Acceptance tests are scheduled for Jan 2002, with a SCIDAR run (Imperial College Gp.) at La Silla in May-June 2002 for 3-4 weeks to detailed comparisons. He pointed out that this is not a robotic instrument, but assuming success, the instrument should be ready for use after this. The CTIO instrument is meant as a "site testing" piece of equipment, and Matthias S. mentioned the probable need for a third copy.
Hugo Schwarz mentioned that one could get some idea of the relative strengths of high and low altitude turbulence layers from a DIMM, by changing exposure times - since slow variations are nearby, fast variations distant. And you could also get the isoplanatic angle, as is done at Paranal by Marc S. Larger aperture DIMM though.
A few other comments made at this time. Matthias S. mentioned measurement of outer scale but no one took the bait. Andrei T. pointed out that for MCAO what was really important was the relative strengths of the high and low turbulence layers and not detailed knowledge of exactly where the layers were. We do need to know time constants, and isoplanatic angle.
Hugo S. also described the CTIO Cloud Camera, which uses a Nikon F/2.8 fisheye lens to image on a TE-cooled SITe 1024 CCD. All commercial, and far superior to CONCAM. It will have filters to also measure light pollution. Near completion.
Briefly mentioned was the CFHT "sky flux" photometer, and a 10 micron fisheye at the Rocky Mountain Observatory ($80K).
And here is a Table of the Status of Generalized Scidar Instruments , from Elena M.This is of course a well known activity for MK, but Riccardo G. made some wise remarks applicable to Chile, and extensible to Mexico and elsewhere presumably. Tread very lightly on local and national sensitivities, get permission for most everything. The Dir. of Archaeology at San Pedro is the person to talk to re possible problems with Quimal.
Hugo Schwarz volunteered to make an email alias for the group at the meeting, plus we should check to see if we should add some more names (I have added the full list from the "NIO-Mexico" sites group).
We decided that since this meeting was such fun we should have more, and decided that 6 months time (ie around May-June 2002) in Ithaca (Cornell) was optimal. By that time Andre E. will have finished the SW USA-Mexico analysis (he could present it there?), CELT may have its CoD studies funded or if not maybe a slightly revised timescale, the meso- and micro- modeling work should have advanced, we should have some results from the MK DIMMs, know hopefully how the MK MM5 predictions work, the Profilometer will have been tested, CTIO should have deployed some weather stations on unsuspecting mountains in the Atacama, and the International Conf. on Light Pollution (in La Serena) will have taken place.
The meso- and micro-scale modeling work, including the UH predictions for MK are thought to be so important that a small meeting of the experts should occur soon. Steve Strom suggested November (later off-line discussions suggested Dec-Jan would be better, by then the UH post-doc and the Elena-Dave post-doc might both have been accreted).
Make contact with the Argentine site-testing group
Draft a summary parameter table of characteristics for each site
Some text (?) on techniques and methods. The CELT "sites chapter" in the GREEN BOOK is an excellent start to this task, and soon will be publically available.
---